blk: Fix lock inversion between ioc lock and bfqd lock

mainline inclusion
from mainline-v5.14-rc1
commit  fd2ef39c
category: bugfix
bugzilla: 185777 https://gitee.com/openeuler/kernel/issues/I4LM14
CVE: NA

------------------------------

Lockdep complains about lock inversion between ioc->lock and bfqd->lock:

bfqd -> ioc:
 put_io_context+0x33/0x90 -> ioc->lock grabbed
 blk_mq_free_request+0x51/0x140
 blk_put_request+0xe/0x10
 blk_attempt_req_merge+0x1d/0x30
 elv_attempt_insert_merge+0x56/0xa0
 blk_mq_sched_try_insert_merge+0x4b/0x60
 bfq_insert_requests+0x9e/0x18c0 -> bfqd->lock grabbed
 blk_mq_sched_insert_requests+0xd6/0x2b0
 blk_mq_flush_plug_list+0x154/0x280
 blk_finish_plug+0x40/0x60
 ext4_writepages+0x696/0x1320
 do_writepages+0x1c/0x80
 __filemap_fdatawrite_range+0xd7/0x120
 sync_file_range+0xac/0xf0

ioc->bfqd:
 bfq_exit_icq+0xa3/0xe0 -> bfqd->lock grabbed
 put_io_context_active+0x78/0xb0 -> ioc->lock grabbed
 exit_io_context+0x48/0x50
 do_exit+0x7e9/0xdd0
 do_group_exit+0x54/0xc0

To avoid this inversion we change blk_mq_sched_try_insert_merge() to not
free the merged request but rather leave that upto the caller similarly
to blk_mq_sched_try_merge(). And in bfq_insert_requests() we make sure
to free all the merged requests after dropping bfqd->lock.

Fixes: aee69d78 ("block, bfq: introduce the BFQ-v0 I/O scheduler as an extra scheduler")
Reviewed-by: NMing Lei <ming.lei@redhat.com>
Acked-by: NPaolo Valente <paolo.valente@linaro.org>
Signed-off-by: NJan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20210623093634.27879-3-jack@suse.czSigned-off-by: NJens Axboe <axboe@kernel.dk>

Conflict: 1. mainline include/linux/elevator.h file change;
          2. mainline block/mq-deadline-main.c file change, now change
             to block/mq-deadline.c;
Signed-off-by: Nzhangwensheng <zhangwensheng5@huawei.com>
Reviewed-by: Nqiulaibin <qiulaibin@huawei.com>
Reviewed-by: NJason Yan <yanaijie@huawei.com>
Signed-off-by: NZheng Zengkai <zhengzengkai@huawei.com>

block/bfq-iosched.c

@@ -2235,9 +2235,9 @@ static bool bfq_bio_merge(struct request_queue *q, struct bio *bio,
 
 	ret = blk_mq_sched_try_merge(q, bio, nr_segs, &free);
 
+	spin_unlock_irq(&bfqd->lock);
 	if (free)
 		blk_mq_free_request(free);
-	spin_unlock_irq(&bfqd->lock);
 
 	return ret;
 }
@@ -5508,14 +5508,16 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq,
 	struct bfq_queue *bfqq;
 	bool idle_timer_disabled = false;
 	unsigned int cmd_flags;
+	LIST_HEAD(free);
 
 #ifdef CONFIG_BFQ_GROUP_IOSCHED
 	if (!cgroup_subsys_on_dfl(io_cgrp_subsys) && rq->bio)
 		bfqg_stats_update_legacy_io(q, rq);
 #endif
 	spin_lock_irq(&bfqd->lock);
-	if (blk_mq_sched_try_insert_merge(q, rq)) {
+	if (blk_mq_sched_try_insert_merge(q, rq, &free)) {
 		spin_unlock_irq(&bfqd->lock);
+		blk_mq_free_requests(&free);
 		return;
 	}
 

block/blk-merge.c

@@ -831,18 +831,15 @@ static struct request *attempt_front_merge(struct request_queue *q,
 	return NULL;
 }
 
-int blk_attempt_req_merge(struct request_queue *q, struct request *rq,
-			  struct request *next)
+/*
+ * Try to merge 'next' into 'rq'. Return true if the merge happened, false
+ * otherwise. The caller is responsible for freeing 'next' if the merge
+ * happened.
+ */
+bool blk_attempt_req_merge(struct request_queue *q, struct request *rq,
+			   struct request *next)
 {
-	struct request *free;
-
-	free = attempt_merge(q, rq, next);
-	if (free) {
-		blk_put_request(free);
-		return 1;
-	}
-
-	return 0;
+	return attempt_merge(q, rq, next);
 }
 
 bool blk_rq_merge_ok(struct request *rq, struct bio *bio)

block/blk-mq-sched.c

@@ -381,9 +381,10 @@ bool __blk_mq_sched_bio_merge(struct request_queue *q, struct bio *bio,
 	return ret;
 }
 
-bool blk_mq_sched_try_insert_merge(struct request_queue *q, struct request *rq)
+bool blk_mq_sched_try_insert_merge(struct request_queue *q, struct request *rq,
+				   struct list_head *free)
 {
-	return rq_mergeable(rq) && elv_attempt_insert_merge(q, rq);
+	return rq_mergeable(rq) && elv_attempt_insert_merge(q, rq, free);
 }
 EXPORT_SYMBOL_GPL(blk_mq_sched_try_insert_merge);
 

block/blk-mq-sched.h

@@ -12,7 +12,8 @@ bool blk_mq_sched_try_merge(struct request_queue *q, struct bio *bio,
 		unsigned int nr_segs, struct request **merged_request);
 bool __blk_mq_sched_bio_merge(struct request_queue *q, struct bio *bio,
 		unsigned int nr_segs);
-bool blk_mq_sched_try_insert_merge(struct request_queue *q, struct request *rq);
+bool blk_mq_sched_try_insert_merge(struct request_queue *q, struct request *rq,
+				   struct list_head *free);
 void blk_mq_sched_mark_restart_hctx(struct blk_mq_hw_ctx *hctx);
 void blk_mq_sched_restart(struct blk_mq_hw_ctx *hctx);
 

block/blk-mq.h

@@ -283,6 +283,17 @@ static inline struct blk_plug *blk_mq_plug(struct request_queue *q,
 	return NULL;
 }
 
+/* Free all requests on the list */
+static inline void blk_mq_free_requests(struct list_head *list)
+{
+	while (!list_empty(list)) {
+		struct request *rq = list_entry_rq(list->next);
+
+		list_del_init(&rq->queuelist);
+		blk_mq_free_request(rq);
+	}
+}
+
 /*
  * For shared tag users, we track the number of currently active users
  * and attempt to provide a fair share of the tag depth for each of them.

block/blk.h

@@ -235,7 +235,7 @@ ssize_t part_timeout_store(struct device *, struct device_attribute *,
 void __blk_queue_split(struct bio **bio, unsigned int *nr_segs);
 int ll_back_merge_fn(struct request *req, struct bio *bio,
 		unsigned int nr_segs);
-int blk_attempt_req_merge(struct request_queue *q, struct request *rq,
+bool blk_attempt_req_merge(struct request_queue *q, struct request *rq,
 				struct request *next);
 unsigned int blk_recalc_rq_segments(struct request *rq);
 void blk_rq_set_mixed_merge(struct request *rq);

block/elevator.c

@@ -353,9 +353,11 @@ enum elv_merge elv_merge(struct request_queue *q, struct request **req,
  * we can append 'rq' to an existing request, so we can throw 'rq' away
  * afterwards.
  *
- * Returns true if we merged, false otherwise
+ * Returns true if we merged, false otherwise. 'free' will contain all
+ * requests that need to be freed.
  */
-bool elv_attempt_insert_merge(struct request_queue *q, struct request *rq)
+bool elv_attempt_insert_merge(struct request_queue *q, struct request *rq,
+			      struct list_head *free)
 {
 	struct request *__rq;
 	bool ret;
@@ -366,8 +368,10 @@ bool elv_attempt_insert_merge(struct request_queue *q, struct request *rq)
 	/*
 	 * First try one-hit cache.
 	 */
-	if (q->last_merge && blk_attempt_req_merge(q, q->last_merge, rq))
+	if (q->last_merge && blk_attempt_req_merge(q, q->last_merge, rq)) {
+		list_add(&rq->queuelist, free);
 		return true;
+	}
 
 	if (blk_queue_noxmerges(q))
 		return false;
@@ -381,6 +385,7 @@ bool elv_attempt_insert_merge(struct request_queue *q, struct request *rq)
 		if (!__rq || !blk_attempt_req_merge(q, __rq, rq))
 			break;
 
+		list_add(&rq->queuelist, free);
 		/* The merged request could be merged with others, try again */
 		ret = true;
 		rq = __rq;

block/mq-deadline.c

@@ -489,6 +489,7 @@ static void dd_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq,
 	struct request_queue *q = hctx->queue;
 	struct deadline_data *dd = q->elevator->elevator_data;
 	const int data_dir = rq_data_dir(rq);
+	LIST_HEAD(free);
 
 	/*
 	 * This may be a requeue of a write request that has locked its
@@ -496,8 +497,10 @@ static void dd_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq,
 	 */
 	blk_req_zone_write_unlock(rq);
 
-	if (blk_mq_sched_try_insert_merge(q, rq))
+	if (blk_mq_sched_try_insert_merge(q, rq, &free)) {
+		blk_mq_free_requests(&free);
 		return;
+	}
 
 	blk_mq_sched_request_inserted(rq);
 

include/linux/elevator.h

@@ -117,7 +117,8 @@ extern void elv_merge_requests(struct request_queue *, struct request *,
 			       struct request *);
 extern void elv_merged_request(struct request_queue *, struct request *,
 		enum elv_merge);
-extern bool elv_attempt_insert_merge(struct request_queue *, struct request *);
+extern bool elv_attempt_insert_merge(struct request_queue *, struct request *,
+				     struct list_head *);
 extern struct request *elv_former_request(struct request_queue *, struct request *);
 extern struct request *elv_latter_request(struct request_queue *, struct request *);